inpainting.h

Go to the documentation of this file.

#ifndef SOPT_TOOLS_FOR_TESTS_INPAINTING_H
#define SOPT_TOOLS_FOR_TESTS_INPAINTING_H
 
#include "sopt/config.h"
#include <random>
#include "sopt/linear_transform.h"
#include "sopt/types.h"
 
namespace sopt {
 
template <typename T>
Vector<T> target(sopt::LinearTransform<Vector<T>> const &sampling, sopt::Image<T> const &image) {
  return sampling * Vector<T>::Map(image.data(), image.size());
}
 
template <typename T>
typename real_type<T>::type sigma(sopt::LinearTransform<Vector<T>> const &sampling,
                                  sopt::Image<T> const &image) {
  auto constexpr snr = 30.0;
  auto const y0 = target(sampling, image);
  return y0.stableNorm() / std::sqrt(y0.size()) * std::pow(10.0, -(snr / 20.0));
}
 
template <typename T, typename RANDOM>
Vector<T> dirty(sopt::LinearTransform<Vector<T>> const &sampling, sopt::Image<T> const &image,
                RANDOM &mersenne) {
  // values near the mean are the most likely
  // standard deviation affects the dispersion of generated values from the mean
  auto const y0 = target(sampling, image);
  std::normal_distribution<> gaussian_dist(0, sigma(sampling, image));
  Vector<T> y(y0.size());
  for (t_int i = 0; i < y0.size(); i++) y(i) = y0(i) + gaussian_dist(mersenne);
 
  return y;
}
 
template <typename T>
typename real_type<T>::type epsilon(sopt::LinearTransform<Vector<T>> const &sampling,
                                    sopt::Image<T> const &image) {
  auto const y0 = target(sampling, image);
  auto const nmeasure = y0.size();
  return std::sqrt(nmeasure + 2 * std::sqrt(nmeasure)) * sigma(sampling, image);
}
}  // namespace sopt
#endif